[postgis-tickets] [PostGIS] #2289: Redesign Geocoder

[PostGIS]

#2289: Redesign  Geocoder
----------------------------+-----------------------------------------------
 Reporter:  woodbri         |       Owner:  robe          
     Type:  enhancement     |      Status:  new           
 Priority:  medium          |   Milestone:  PostGIS Future
Component:  tiger geocoder  |     Version:  2.0.x         
 Keywords:                  |  
----------------------------+-----------------------------------------------

Comment(by robe):

 Which tiger year is that?  I have in featnames (I think I have 2012
 loaded) ( 564,976 records with non-empty fullnames for MA) granted some
 might not be things usually addressable.  so I would expect your table to
 be a little bigger. Maybe not. Anyway I like the idea of a sideline table
 rather than adding a column to a tiger table so I might start with that
 approach.

 The main reason I haven't denormalized it is I'm not convinced it makes
 speed faster for PostgreSQL (for MySQL or SQLite Of course :) ). Although
 there is more joining - PostgreSQL is pretty good at that.  What you get
 with multiple tables is possibility of using multiple disk seeks (since
 you can have the different sets on different disks) and possibility of
 loading more records in shared memory since you won't need to duplicate.

 So before going that route, I'd have to benchmark how the loss in the
 aforementioned is improved with a single table set. Granted it will make
 query easier to understand to others, but that's the least of my problems
 :)

 Aside from that, disk space is a big concern for me mostly because
 1) I have to worry about cloud disks which are often limited in size and
 scaled based on processor (except for Amazon disks which you pay for by
 the super suckiness).
 2) It takes longer to copy my tables from one install to another the
 larger the dataset.

-- 
Ticket URL: <http://trac.osgeo.org/postgis/ticket/2289#comment:5>
PostGIS <http://trac.osgeo.org/postgis/>
The PostGIS Trac is used for bug, enhancement & task tracking, a user and developer wiki, and a view into the subversion code repository of PostGIS project.